AI Search Is Turning Legal SEO Into a Boardroom Decision About How Firms Compete

Legal marketing used to be a contest over keywords and map-pack rankings, a technical job that lived a floor below the people running the firm. AI search and Google's recent core updates are pulling it upstairs. To get surfaced now, a firm has to give search systems real evidence of its expertise, its reputation, and its standing in the community, which turns a task that used to sit with an SEO into a decision about how the firm competes. Firms that treat this as a positioning problem, rather than a ranking one, are the firms getting found.

Daniel Epstein is the Founder of AiTorney and a Fractional CMO who has spent decades building and selling marketing agencies, including his current legal-focused shops, Stratophi and the Raleigh SEO Team. His background runs through mergers and acquisitions, where he watched a company's value climb once its online presence did, and that deal-side lens shapes how he approaches search for law firms. He treats it less as a marketing channel and more as a lever on how the whole firm is perceived.

That reframing is the heart of his pitch. Ranking a firm is only the entry point, and the real work is deciding how it should be seen. "Instead of just thinking about ranking somebody, now what we're really thinking about is positioning somebody," says Epstein. "It's C-level thinking." Positioning, in his framing, means bringing search into the same conversations as strategy and operations, and it starts with what the firm can actually prove about itself.

From content to credibility

Each earlier phase of SEO rewarded a different thing, Epstein says: first just having a website, then keywords, then the map pack, then content. The current phase rewards credibility. Google's core updates lean on what it calls E-E-A-T, for experience, expertise, authoritativeness, and trustworthiness, and legal sits squarely in the "your money or your life" category the company scrutinizes hardest. "Now it's not just a matter of us being SEO practitioners," says Epstein. "It's really more about being a strategic partner with a firm." The signals that count have widened well past the page.

He frames the change as good for consumers and for firms doing genuine work. A year ago, he says, Google paid no attention to whether a firm sponsored a little league team or showed up for its community, and now that kind of real involvement registers as evidence of trust. That rewards firms with a real story and exposes ones with a thin one, which is why he leans on the trust firms build over years rather than anything a landing page can manufacture. "Let your personality shine and show the community who you are," he says.

A level field for small firms

The shift has opened a window for the firms that usually get outspent. Most law firms in the United States are small, Epstein notes, while most of the firms pouring money into AI are large, and the distance between them has never mattered less. "The playing field has never been as level as it is right now," says Epstein. A small firm with a clear position, in his account, can climb into local rankings within a couple of months and then start appearing in AI answers on the strength of strategy rather than spend.

His confidence on that point is high, and the mechanism is the same one a small firm competing on reputation can use: hand the systems credible, specific evidence and let them do the sorting. What large budgets bought in the old model, saturation and paid placement, carries less weight when an assistant is weighing which firm best answers a searcher's question. The advantage moves to whoever supplies the clearest, most verifiable case.

Positioning, not a plugin

In practice, the work starts with diagnosis. Epstein's team sends a new firm a detailed survey, maps its strengths, weaknesses, opportunities, and threats, then builds a content strategy specific to the market and the niche. A family-law practice and a personal-injury firm get handled differently, because the competitive picture and the buyer are different. "It's not a one-size-fits-all solution," says Epstein. "Every market, every niche is different."

From there it becomes what he calls trophy content, surfacing a firm's real awards, wins, and experience so both search systems and buyers can see it. He's clear that the approach only works when the firm has something to show, since E-E-A-T can't be fabricated, and a firm with no community involvement and no interest in differentiating itself can't be dressed up as the most trustworthy option. The builds are modernizing alongside it, moving toward faster static sites as he bets that websites are on their way to becoming apps, all of it aimed at shaping how models surface the firm.

Where AI goes wrong

For all his enthusiasm, Epstein thinks far more people are using AI badly than well. The worst habit is generating and publishing AI content wholesale, since the output is only as good as what it's fed, and most of what gets pushed out is filler. "The biggest mistake is that there's so much AI fluff that's so meaningless it has very little value," says Epstein, who adds that the flood of it actually makes his job easier by lowering the bar around him.

He points to how confidently wrong these systems can be. Affiliate marketers with no real SEO knowledge pump out misinformation, and he cites a test in which Search Atlas invented a fake state called "Smackachusetts" and got AI systems to describe it in convincing detail. That, to him, is exactly why verifiable expertise matters, and why content at scale without a credible source underneath it becomes a liability rather than an asset. Google's own incentive is to surface the best possible answer, he notes, which punishes the fluff over time.

Keeping the agents on a leash

The governance question gets sharper in law, where professional rules limit what a firm is allowed to claim. Epstein learned it the hard way about two and a half years ago, when one of his early agent builds for a law firm went off the rails, pulling case information and effectively starting to practice law and make prohibited claims. "It went off the guardrails," says Epstein. "It was literally pulling information on cases and practicing law," the kind of thing that draws a bar association's attention fast.

His team took those agents down, learned to build guardrails, and changed how they run them. Agents now live on contained servers rather than personal machines and stay under active monitoring, because moving work to agents only pays off when someone stays accountable for what they do. He points to the jailbreak problems the AI labs themselves discuss, and to companies that restrict what their own developers can run locally, as reasons to treat containing the agents as seriously as deploying them. "You really have to know how to contain your agents," he says, "because these things are crazy."